Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Oct 22.
Published in final edited form as: J Physiol. 2024 May 15;602(20):5163–5178. doi: 10.1113/JP285120

Single-cell ionic current phenotyping elucidates noncanonical features and predictive potential of cardiomyocytes during automated drug experiments

Alexander P Clark 1, Siyu Wei 2, David J Christini 1,2, Trine Krogh-Madsen 3,4
PMCID: PMC11493530  NIHMSID: NIHMS1989974  PMID: 38747042

Abstract

All new drugs must go through preclinical screening tests to determine their proarrhythmic potential. While these assays effectively filter out dangerous drugs, they are too conservative, often misclassifying safe compounds as proarrhythmic. In this study, we attempt to address this shortcoming with a novel, medium-throughput drug-screening approach: we use an automated patch-clamp system to acquire optimized voltage clamp (VC) and action potential (AP) data from human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) at several drug concentrations (baseline, 3x, 10x, and 20x the effective free plasma concentrations). With our novel method, we show correlations between INa block and upstroke slowing after treatment with flecainide or quinine. Additionally, after quinine treatment, we identify significant reductions in current during voltage steps designed to isolate If and IKs. However, we do not detect any IKr block by either drug, and upon further investigation, do not see any IKr present in the iPSC-CMs when prepared for automated patch experiments (i.e., in suspension) – this is in contrast to similar experiments we have conducted with these cells using the manual patch setup. In this study, we: 1) present a proof-of-concept demonstration of a single-cell medium-throughput drug study, and 2) characterize the noncanonical electrophysiology of iPSC-CMs when prepared for experiments in a medium-throughput setting.

Graphical Abstract

graphic file with name nihms-1989974-f0011.jpg

Abstract Figure: In this study, we use a medium-throughput automated patch-clamp system to acquire action potential (AP) and complex voltage clamp (VC) data from single human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) at multiple drug concentrations. A correlation between AP upstroke and peak INa was identified and drug-induced changes in additional ionic currents found. We also characterize the substantially altered physiology of iPSC-CMs when patched in an automated system, suggesting the need to investigate differences between manual and automated patch experiments.

1. Introduction

All new drugs must go through preclinical screening tests to determine their proarrhythmic potential. These tests include animal long QT studies and a drug’s interaction with the human ether-a-go-go-related gene (hERG) ion channel (U.S. Food and Drug Administration, 2005). While these methods have proven accurate at detecting cardiotoxic drugs, they are too conservative, regularly misclassifying safe and potentially beneficial drugs as proarrhythmic (Johannesen et al., 2014). As such, there is a need to develop preclinical screening approaches that maintain the high sensitivity of current methods, while improving specificity, leading to more safe and effective drugs reaching clinical trials.

In 2014, an international group of regulatory agencies started the Comprehensive in Vitro Proarrhythmia Assay with the aim of developing a new preclinical drug screening approach that included (Sager et al., 2014): 1) a focus on several key cardiac ionic currents (in addition to hERG), 2) the use of in silico drug screening tests with computational action potential (AP) models, and 3) in vitro studies of drug effects on human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs). The focus on multiple ion channels and use of human-derived myocyte assays is expected to provide greater mechanistic insight and improve the physiological relevance of findings when compared to traditional single-channel (e.g., hERG) expression system studies. However, while meaningful progress has been made on all three of these components, in vitro iPSC-CM studies (Phase 3) have been limited by the immature and heterogeneous phenotype of these cells (Goversen et al., 2018).

The heterogeneous phenotype of iPSC-CMs has led to confounding drug results with multiple labs producing opposite outcomes to the same drug treatment (Blinova et al., 2019). Such heterogeneity has motivated work into developing iPSC-CM protocols that increase maturation and reduce heterogeneity (Lyra-Leite et al., 2022) — the impressive rate of new publications in this space makes it difficult for users to readily adopt the latest protocol. This can lead to labs selecting and sticking with a fit-for-purpose differentiation approach that works for their specific studies, providing one explanation for lab-to-lab differences. This variation in iPSC-CM phenotype is, almost certainly, due in part to disparate ionic current expression levels from cell-to-cell and lab-to-lab. Unfortunately, fully characterizing the ionic current expression levels of iPSC-CMs in a lab is unrealistic for two reasons: 1) voltage clamp protocols are laborious, and 2) there is no guarantee that subsequent differentiation batches have the same expression levels.

In previous work (Clark et al., 2022, 2024), we have shown how an optimized voltage clamp (VC) protocol can be used to characterize the ionic current expression of individual cells, explain AP morphology, and provide mechanistic insights into drug-induced changes. To the best of our knowledge, the findings from these studies were the first to show how acquiring multi-channel VC and AP recordings from the same cells can provide a detailed picture of iPSC-CM electrophysiology. While this recent work demonstrated promise as a method to improve the interpretation of iPSC-CM-based cardiotoxicity tests, the approach was limited by the low-throughput nature of the manual patch technique.

In this study, we set out to build on Clark et al. (2022, 2024) by implementing a drug screening pipeline that uses single-cell ionic current phenotyping to interpret changes in both AP and rich VC data acquired from iPSC-CMs using the Nanion Patchliner, an automated patch-clamp system (Becker et al., 2020). Here, the automated system provides three improvements upon our previous manual setup: 1) it is considerably higher throughput, 2) there is no need for patch clamp expertise to conduct the experiments, and 3) it is easy to apply multiple concentrations of a drug in a timely manner. We use both flecainide and quinine (both strong INa and IKr blockers) to demonstrate the utility of this novel approach. We show correlations between INa block and upstroke slowing. But, from these same cells, we do not see any evidence of AP prolongation or IKr block after drug treatment. Upon further investigation, we do not see evidence of IKr in our iPSC-CMs when they are prepared (in suspension) for testing in the automated patch machine, despite us previously validating the presence of IKr in cells from the same cell line during manual patch experiments (cell-attached). In all, this study achieves two important steps towards realizing the potential of a high-throughput fully-automated iPSC-CM electrophysiology drug screening platform:

  1. Demonstrates, as a proof of concept, the ability to acquire AP and rich multi-channel VC data from the same cell at several drug concentrations.

  2. Characterizes the noncanonical electrophysiology of iPSC-CMs in the automated patch system, motivating the need and providing a direction for further optimization of cell lines to be used in high-throughput systems.

2. Methods

2.1. Ethical Approval

Human iPSC-CMs were purchased from FUJIFILM Cellular Dynamics, Inc and Greenstone Biosciences. Both cell lines were derived from consenting donors, in accordance with the Declaration of Helsinki and journal policy.

2.2. Experimental design

We used the four-channel Nanion Patchliner with a Dynamite8 add-on to conduct these experiments (Becker et al., 2020). We developed a custom protocol in PatchControl and PatchMaster that provided near-complete automation of every experimental step after iPSC-CM dissociation.

Figure 1 shows the acquisition protocol. The Patchliner first acquired optimized VC data (see 2.2) at baseline, before then switching to current clamp mode where we acquired >1min of dynamically clamped AP data (see 2.5). The time sensitive nature of the protocol makes it important to quickly identify suprathreshold stimuli for all channels. Therefore, we choose a large stimulus to ensure AP generation in all channels for the duration of the pacing protocol. This results in a few cells being overstimulated and having a very polarized overshoot (AP peak >50mV, Figure 3A). The Patchliner added 3xEFPC of drug (DMSO control, flecainide, or quinine) to the cells while APs were continuously acquired. APs acquired from >75s after drug application were compared to pre-drug data. This timepoint was selected to provide enough time to observe drug effects, while not prolonging the total protocol beyond the period where the cells remain viable. Following AP acquisition, the Patchliner switched back to VC mode and acquired VC data at 3xEFPC.

Figure 1: Cardiotoxicity screening experimental setup.

Figure 1:

Open in a new tab

We use the Nanion Patchliner to acquire both dynamically clamped AP (A, see 2.5 for details) and optimized VC data (B) from iPSC-CMs at baseline, after the addition of 3x, 10x, and 20xEFPC, and after wash (C).

Figure 3: After injection of hyperpolarizing currents, iPSC-CM APs were short in duration and electrophysiologically heterogeneous.

Figure 3:

Open in a new tab

(A), An example AP recording from each cell used in this study. Distributions of APD90 (B), MDP (C), and dV/dtmax (D) for the cells displayed in A. All APs were recorded with constant hyperpolarizing, and dynamically clamped IK1 and seal-leak currents.

This is the pattern we used to acquire VC and AP data at baseline and after each drug/wash addition. We conducted this study with flecainide (n=12) and quinine (n=14). These drugs were selected because they are well-characterized, known to be proarrhythmic, and block multiple key cardiac ionic currents. Studies with these drugs were compared to cells treated with a DMSO (n=7) control solution. One of the DMSO cells was excluded from AP analyses, because the system errored while collecting AP data. We included all cells that survived to the end of the experimental protocol and had a measurable upstroke.

2.3. Developing an optimized VC protocol for rapid ionic current phenotyping

We optimized a VC protocol for rapid ionic current phenotyping (RICP) using a genetic algorithm (GA) and the Kernik+artifact iPSC-CM cardiomyocyte model, as described previously (Clark et al., 2022). The VC protocol for the RICP approach consists of multiple steps, each optimized to maximize the contribution from a different ionic current. We made two modifications to the approach in this study:

  • We added INaL (Tomek et al., 2019) to the Kernik+artifact model (Kernik et al., 2019; Lei et al., 2020) that we used in Clark et al. (2022). The conductance of INaL was scaled (while holding baseline gNa constant) to match the ratio of peak gNa:gNaL in the Tomek et al. (2019) model. This current was added so we could include an INaL-isolating segment to the VC protocol.

  • Segments in the VC protocol were limited to just VC steps of constant voltage — previously, segments could be steps or ramps. We made this simplifying modification because we did not see substantial improvements (during in silico simulations) in current isolation when we included ramps.

The GA-optimized IKr, If, and IKs steps (Figure 2) were designed by the algorithm to maximize the respective current (e.g., IKr, If, or IKs) within 1ms after a voltage step. Due to the sensitivity of the cell to experimental artifact parameters (e.g., access resistance) just after voltage steps, we chose to instead consider current measurements just before the step — these timepoints isolate each current well and are free from transient voltage step-induced artifact.

Figure 2: The optimized VC protocol that we acquire from every cell.

Figure 2:

Open in a new tab

The arrow points to the GA-designed isolating segment for INa. After acquiring data, we determined that a different segment more consistently elicited large INa transients (labeled Na), and was used for further analyses throughout this study.

After acquiring data, we determined that our model-designed INa-isolating segment (Figure 2, arrow) did not always generate a large INa peak. As such, we chose to also investigate a different INa-isolating segment (labeled Na) that more consistently elicited a large INa transient.

The isolating segments of most currents (INa, IKr, Ito, IK1, If, and IKs) are very similar to the protocol designed in (Clark et al., 2022). There are two notable differences:

  • This protocol includes an INaL-isolating segment

  • The ICaL-isolating segment is different. Instead of stepping up to a ramp (see Clark et al. (2022)), the new VC protocol takes a brief step up to 3mV to activate calcium channels and then steps down to a large negative potential to increase the driving force for calcium.

2.4. iPSC-CM culture and dissociation

iCell cardiomyocytes (11713 C1105) derived from a healthy African American female were obtained from FUJIFILM Cellular Dynamic, Inc and were thawed and plated with iCell Cardiomyocytes Plating medium and maintained in iCell Cardiomyocytes maintenance medium as instructed in iCell cardiomyocytes user’s guide.

Human iPSC-CMs derived from a healthy white female individual (SCVI-480CM) were obtained from Greenstone Biosciences, Inc. Frozen vials of iPSC-CMs were thawed and cultured as a monolayer in one well of a 6-well plate precoated with 1% Matrigel and supplemented with RPMI media (Fisher/Corning 10–040-CM) containing 5% FBS (Gibco 16000069) and 2% B27 (Gibco A1895601). Cells were placed in an incubator at 37°C, 5% CO2, and 85% humidity for 7–14 days. On the day of the experiment, cells were exposed to 250μL of Accumax and the enzymatic reaction was stopped after 20–40 minutes with the addition of 1.5mL of extracellular solution. The cells were stored in a 4°C refrigerator for 15 minutes after dissociation. Cells were passed through a 100μm filter before use to remove the large clumps of cells that remain after dissociation.

2.5. Patch clamp experiments

Whole cell rupture patch experiments were conducted at room temperature (22–25°C) using standard medium resistance NPC-16 chips. The extracellular solution contained 130 mM NaCl, 4mM KCl, 10 mM CaCl2, 1mM MgCl2, 5mM D-glucose monohydrate, 10 mM HEPES, and with pH adjusted to 7.4 with NaOH. The intracellular solution contained 110 mM KF, 10 mM NaCl, 10 mM KCl, 10 mM EGTA, 10 mM HEPES, and with pH adjusted to 7.2. Fresh drug solutions were made the morning of each experiment. Series resistance compensation, including supercharging (see Lei (2020) for details), was set to 70% for VC experiments.

The liquid junction potential was calculated to be 10mV using the Nernst–Planck equation (Marino et al., 2014) and LJPcalc software (https://swharden.com/LJPcalc). This was corrected for during both VC and dynamic clamp the experiments.

2.6. IK1 dynamic clamp and other injected currents during AP recordings

In current clamp mode, cells were very depolarized (>−15mV) and did not beat spontaneously, which is similar to previous results of iPSC-CMs in the automated patch system (Goversen et al., 2018). To recover a resting membrane potential of <−70mV, cells were simultaneously injected with: 1) an IK1 dynamic clamp current, 2) dynamic clamp seal-leak compensation current, and 3) a constant current to hyperpolarize the cell towards a negative resting membrane potential (Becker et al., 2020). The IK1 current (Ishihara et al., 2009) has previously been shown to improve the AP morphology of iPSC-CMs for use in drug cardiotoxicity studies (Goversen et al., 2018). The seal-leak compensation current was previously implemented (Ahrens-Nicklas and Christini, 2009) during patch clamp experiments to counter the effects of seal-leak between the pipette tip and cell membrane. Due to the very (relative to iPSC-CMs during manual experiments) depolarized potential of these cells, we also had to inject a constant hyperpolarizing current. This hyperpolarizing current was required during previously published experiments with iPSC-CMs and primary cardiomyocytes used in automated patch systems (Becker et al., 2020; Seibertz et al., 2022).

2.7. Software and simulations

Simulations were performed in Myokit v1.33.7 (Clerx et al., 2016). Additional analysis, including linear least squares regressions, was done in Python using NumPy v1.21.6 and SciPy v1.7.3 (Virtanen et al., 2020).

All code has been made publicly available on GitHub at: https://github.com/Christini-Lab/automated-ipsc-cardiotoxicity.git

3. Results

3.1. iPSC-CM APs are heterogeneous and require strong hyperpolarizing current before pacing

The iPSC-CMs used in this study were purchased from two vendors (Greenstone and FUJIFILM; see Methods for more details). Cells had an average capacitance and series resistance (mean±SD) of 25.2±11.6pF and 7.9±5.0MΩ. At rest, all cells in each group were substantially depolarized (>−50mV) and did not produce spontaneous APs. This is different from spontaneous manual patch-clamp recordings made with the Greenstone cells (Clark et al., 2022). Due to the depolarization of the cells in this study, we inject three hyperpolarizing currents (leak compensation, IK1 dynamic clamp, and a constant hyperpolarizing current) to maintain a resting membrane potential below −70mV (see Methods for more details). After current injection and MDP stabilization, cells were paced at 1Hz — the resultant APs were short in duration and electrophysiologically heterogeneous (Figure 3A), with large variations in APD90 (50 ± 18ms), MDP (−92 ± 6mV), and dV/dtmax (87 ± 44V/s) reported as mean ± SD (Figure 3B-D).

3.2. Ionic current conductances vary widely

In addition to APs, we also obtained optimized VC data recordings for each cell (Figure 4A). This optimized VC protocol is less than 10s in duration and was designed to isolate each of eight key ionic currents (INa, INaL, IKr, ICaL, Ito, IK1, If, IKs) at brief time spans throughout the recording. These time spans represent model-predicted windows of an individual current’s maximum contribution to the total measured current (Iout), but do not entirely isolate each current of interest (see Methods for more details). Figure 4A illustrates the heterogeneity in ionic currents of cells across the entire VC protocol. Figure 4B-C provides an example of the range of responses seen from individual cells during the INa- and If-isolating segments, both currents that we expect to be present in these cells (Clark et al., 2022). Figure 4B shows how some cells produce large INa transients (blue) while others do not (orange). Similarly, in Figure 4C, a step to −120mV shows that some cells (e.g., blue) produce a gradually decreasing Iout measure consistent with If and others do not (e.g., orange).

Figure 4: Responses of iCell and Greenstone iPSC-CMs to voltage command were heterogeneous.

Figure 4:

Open in a new tab

Iout responses to the optimized VC protocol (A), and zoomed views of INa- (B) and If-isolating (C) segments. The blue tracing (B, C) shows evidence of INa and If, while the orange does not.

Seal-leak current, which we cannot easily correct for in iPSC-CMs (Clark et al., 2023), is also present and varying in these cells — we calculated (using estimated Rseal measures) that seal-leak current contributes −4.8±2.8A/F at −80mV (reported as mean±SD).

3.3. Flecaininde and quinine reduce upstroke velocity

AP and optimized VC data were acquired from iCell and Greenstone cells before and after treatment with increasing concentrations of a DMSO control or drug (flecainide or quinine) solution — drug solutions were increased from 3xEFPC to 20xEFPC before being washed out (see Methods for more details). Figure 5A-B show the AP upstroke of two example cells treated with increasing concentrations of flecainide or quinine. Both example cells show a gradual reduction in upstroke velocity with increasing drug concentration, followed by an increase in upstroke velocity after wash. We observe significant differences in the percent change in upstroke velocity for both flecainide- (n=10) and quinine-treated (n=8) cells when compared to the DMSO group (n=6) and observe no difference in upstroke velocity after wash off. This is consistent with block of INa, a known target of both drugs (Crumb et al., 2016).

Figure 5: Flecainide and quinine reduce dV/dtmax.

Figure 5:

Open in a new tab

Example 1Hz-paced AP upstrokes for a flecainide (A) and a quinine-treated (B) cell from baseline (black) to 3xEFPC, 3x to 10x, 10x to 20x, and 20x to wash. Percent change of dV/dtmax in flecainide-treated (C) cells (red) compared to DMSO control (black). Significant differences were observed when comparing DMSO to flecainide at baseline to 3xEFPC (p=.00225), 3x to 10x (p<.001), and 10x to 20x (p<.001). Percent change of dV/dtmax in quinine-treated (D) cells (red) compared to DMSO control (black). Significant differences were observed when comparing DMSO to quinine at baseline to 3xEFPC (p=.00210), 3x to 10x (p<.001), and 10x to 20x (p<.001).

3.4. INa block explains reduction in upstroke velocity

We used the INa-isolating segment of the VC protocol to determine if, as hypothesized above, INa is blocked by quinine and flecainide. INa was elicited during the VC protocol by stepping from −116mV to −31mV. Figure 6A-B shows traces from a flecainide- and a quinine-treated cell during the INa-isolating segment of the VC protocol. As expected (Crumb et al., 2016), we observe a decrease in peak INa with increased flecainide and quinine concentrations, followed by some amount of recovery after wash (gray). When compared to DMSO control cells, we observe a significant reduction in INa at 10x and 20xEFPC for both flecainide- and quinine-treated cells (Figure 6C-D).

Figure 6: Flecainide and quinine reduce current during INa-isolating segment.

Figure 6:

Open in a new tab

Increasing concentrations of flecainide (A, C) and quinine (B, D) lead to a decrease in peak INa, with some amount of recovery after DMSO wash (gray). The gray points in (C, D) are from the same control experiments, where cells were treated with a DMSO control solution. Significant difference(s) in INa change was not observed when comparing DMSO to flecainide at 3x (p=.363), but were at 10x (p=.00386), and 20xEFPC (p=.0120). Significant difference(s) in INa change was not observed when comparing DMSO to quinine at 3x (p=.174), but were at 10x (p=.00112), and 20xEFPC (p=.00971).

INa conductance is one of the main determinants of upstroke velocity. Because we have both AP and VC data at five concentrations for each cell, we can plot the upstroke velocity against peak INa currents to investigate the correlation between these two measurements (Figure 7). As expected, regression analyses for flecainide- (R=−0.79, p<.001) and quinine-treated (R=−0.78, p< .001) cells shows there is a significant relationship between dV/dtmax and INa. This correlation supports the causal effect of INa block on dV/dtmax slowing. This analysis illustrates the power and benefits of acquiring AP and ionic-current optimized VC data at multiple concentrations with an automated patch-clamp system.

Figure 7: Reduction in INa correlates with upstroke slowing.

Figure 7:

Open in a new tab

Peak INa plotted against dV/dtmax for flecainide- (A) and quinine-treated (B) cells. Data points are taken from data at baseline, after 3x, 10x, and 20xEFPC drug application, and after a wash step.

3.5. Lack of IKr in iPSC-CMs explains no change in APD90

Both flecainide and quinine are known to block IKr (Crumb et al., 2016), carry risk of proarrhythmia (Woosley et al., 2021), and expected to prolong the AP duration. The cells used in this study, however, do not show any change in APD90 or Iout during the IKr-isolating segment after treatment with either flecainide ((Figure 8A, C)) or quinine (Figure 8B, D).

Figure 8: No significant change in APD90 or in the IKr-isolating segment after flecainide and quinine treatment.

Figure 8:

Open in a new tab

Change in APD90 from baseline for flecainide- (A) and quinine-treated (B) cells (red) compared to DMSO control (black). Change in current during the IKr-isolating segment for flecainide- (C) and quinine-treated (D) cells compared to DMSO control. There was no significant difference when comparing DMSO to flecainide- or quinine-treated cells at 3x, 10x, or 20xEFPC in any panel. The p-values (comparing at 3x, 10x, and 20xEFPC) for these studies were .594, .518, and .256 for (A); .962, .833, and .619 for (B); .562, .419, and .267 for (C); .719, 0.251, and 0.075 for (D).

In a recent study (Clark et al., 2022), we collected manual-patch AP recordings and VC data with a very similar IKr-isolating segment from Greenstone iPSC-CMs. We observed quinine-induced APD90 prolongation and reduction in current during the IKr-isolating segment. Figure 9A shows an example cell from that study with evidence of IKr block from pre- to post-quinine treatment. In contrast, a quinine-treated cell in the automated patch-clamp system collected as a part of the current study (Figure 9B) shows no change in current after additions of increasing quinine. Baseline traces from the IKr-isolating segment in cells from the current study (Figure 9C) show little evidence for the presence of an IKr-like tail current. The blue and orange traces highlighted in Figure 9C exemplify this lack of tail current.

Figure 9: Cells from the same vendor show IKr presence and block during manual patch, but not during automated.

Figure 9:

Open in a new tab

(A) Current traces from the IKr-isolating segment of a different optimized VC protocol before and after drug treatment shows reduction in current (Clark et al., 2022). (B) Example quinine-treated cell from this study shows no change after quinine treatment. (C) Current traces from IKr-isolating segment of cells from the current study.

3.6. Quinine causes changes to total current consistent with If and IKs block

We observe significant differences in the IKs- and If-isolating segments with increasing concentrations of quinine. At 20xEFPC, we detect a significant increase in total current during the If-isolating segment. This is consistent with our recent observation of a quinine block of inward If current at hyperpolarized negative voltages (Clark et al., 2022). Additionally, we observe significant reductions in outward current during the IKs-isolating segment with increasing quinine concentration. Taken together, these findings agree with previous work finding that quinine is a promiscuous drug that strongly blocks several potassium currents (Clark et al., 2022; Crumb et al., 2016).

4. Discussion

Building on our recent study that used manual patch clamp (Clark et al., 2022), here, we demonstrate the benefits and limitations of using automated patch clamp (specifically the Nanion Patchliner system) for drug cardiotoxicity screening. By collecting both AP and rich VC data from the same cells, we are able to correlate drug-induced AP upstroke slowing to INa block, suggesting a causal relationship. We also used the VC protocol to show that IKr is not measurable to any significant degree in our iPSC-CMs when conducting automated patch experiments. Ultimately, these results show both the tantalizing potential of using cardiomyocytes in automated patch-clamp systems, and the limitations that need to be addressed before its broad acceptance as a tool in drug cardiotoxicity screening.

4.1. APs are brief, heterogeneous, and depolarized in the automated patch setting

The iPSC-CMs used in this study are heterogeneous and their APs are short in duration. Both cell-to-cell differences in ionic current expression (Clark et al., 2024) and exogenous seal-leak current (Clark et al., 2023) can lead to substantial variation in AP morphology. Additionally, the unbiased selection of cells by the automated system makes it possible that a few non-cardiomyocytes could have been selected (as iPSC-CM samples cannot be completely pure), which could be investigated in future studies with large sample sizes. Given the injection of seal-leak compensation, we expect that most of the heterogeneity in this study is due to variation in ionic current expression. Compared to spontaneous APs collected after a biased selection process with manual patch (see accompanying review for more details), the data in this study is less varying (coefficient of variation in resting membrane potential of 36% in this study vs 55% in Clark et al. (2024)) — this is likely because the injected compensatory currents lead to a more uniform morphology. These compensatory currents (IK1/seal-leak dynamic clamp and constant hyperpolarizing current) also contribute to the very short AP duration observed in the cells from this study. Compared to studies that include only IK1 dynamic clamp (Fabbri et al., 2019; Verkerk and Wilders, 2021), the hyperpolarizing current shortens the AP duration and creates an AP morphology that is not physiologically realistic (Goversen et al., 2018). This hyperpolarizing current is required because the cells are depolarized to potentials >−15mV where IK1 is not activated (Fabbri et al., 2019).

Others who acquired APs from iPSC-CMs with the Nanion Patchliner while applying IK1 dynamic clamp also report shortened APD90 values, though the average duration (118±21ms) is longer than we see in our study (Goversen et al., 2018). iPSC-CM and primary pig cardiomyocyte APs acquired using the SyncroPatch, which is not set up for IK1 dynamic clamp but requires a constantly injected hyperpolarizing current, also show abbreviated APs with similar duration to the cells in this study (Seibertz et al., 2022).

The need for injecting several hyperpolarizing currents is due to the substantially depolarized potential of cells tested in the automated setup. This depolarized potential could be due, at least in part, to lower quality seals and the lack of IKr in iPSC-CMs when prepared for automated patch experiments — we (Clark et al., 2024) and others (Doss et al., 2012) have shown that IKr can play a critical role in establishing the resting membrane potential. When cells are prepared for automated patch experiments, they appear to be largely lacking the repolarizing potassium currents that are present during manual patch experiments leading to this depolarized potential (see Sec. 4.3 for further discussion of this). These limitations represent a trade-off between throughput and physiological relevance that must be considered when deciding whether to conduct manual or automated patch experiments with iPSC-CMs.

4.2. Acquiring AP and rich VC data from the same cells

This is the first study we are aware of that attempts to correlate VC data to AP features acquired from the same cell using an automated patch-clamp system. Using RICP, as described in Clark et al. (2024), we collected information about several currents with a simple <10s protocol. This protocol is much briefer than more involved protocols that are typically used to characterize ionic currents in iPSC-CMs (Ma et al., 2011). Such protocols provide more detailed information about current gating kinetics, but the acquisition time is prohibitively long for drug studies that require readouts at multiple concentrations.

The relationship we demonstrate between INa and upstroke velocity provides an example of how RICP can be used to provide a mechanistic explanation for changes seen in AP morphology. While it is known that flecainide and quinine block INa (Crumb et al., 2016) and slow upstroke velocity (Borchard and Boisten, 1982; Sheldon et al., 1995), this is the first study that we know of to identify INa block and upstroke slowing in the same cells within an automated patch system.

Overall, we show which of the ionic currents and AP features are susceptible to drug treatment in the automated patch-clamp system. Specifically, we demonstrate that the system can be used to investigate the effects of INa-blocking drugs (class I antiarrhythmics) on AP features and VC responses, and we show that strong, multi-channel blockers of potassium currents can be identified using the optimized VC protocol.

4.3. iPSC-CMs do not show evidence of IKr during automated patch clamp

Using the optimized VC protocol and RICP, we did not see IKr block after the addition of flecainide or quinine, two strong IKr-blocking drugs (Crumb et al., 2016); nor did we see any AP prolongation. This is in contrast to iPSC-CM experiments we conducted using quinine during manual patch experiments (Clark et al., 2022). Others have reported a similar lack of IKr in the automated patch system (Li et al., 2021) and a recent review stated that the authors were unaware of any group that has reported IKr in iPSC-CMs when using automated patch with cells in suspension (Ismaili et al., 2023). Most automated patch experiments have been conducted at room temperature, a condition that leads to a reduced peak IKr (Lei et al., 2019), and may contribute to the difficulty in identifying the current in iPSC-CMs. However, a group recently showed a very small IKr conductance in their iPSC-CMs during APC experiments that could be substantially enhanced by using a solution with Cs+ ions in place of K+ (Bloothooft et al., 2024). Interestingly, we and others do see IKr when conducting studies using hERG expression line cells in the automated patch system (Polonchuk, 2012; Bloothooft et al., 2024), indicating this issue is not ubiquitous across all cell types.

During manual patch, cells are attached to a cover slip during the experiments, while automated patch experiments like those reported here require that cells be in suspension. The automated patch experiments outlined in this study require dissociation of cells just before the experiments. Given this setup, we hypothesize that the dissociating agent may cause a decrease in transmembrane ion-conducting IKr channels, possibly through disruption of hERG-integrin linking, or that iPSC-CMs do not express hERG at measurable levels when in suspension.

4.4. Limitations and future directions

This study demonstrates the promise of acquiring AP and rich VC data from iPSC-CMs using the Nanion Patchliner, but also identifies a few key limitations that need to be addressed before the method can be used more widely.

First, the community needs to determine why iPSC-CMs do not generate IKr in measurable levels during automated patch experiments. Is the channel not present or not functional? Can IKr be increased in iPSC-CMs by hERG channel activators (Bloothooft et al., 2024) and still allow testing of hERG block by other compounds? Finding a solution to the lack of IKr issue has far-reaching implications for both basic science research and industrial drug cardiotoxicity studies.

Second, the depolarized potential of iPSC-CMs in the automated system requires the injection of a constant hyperpolarizing current that is not typically required during dynamic clamp manual patch experiments. We believe that identifying differences in functional ionic current densities and experimental artifact contributions between iPSC-CMs tested in automated and manual patch experiments will provide insight into the cause of this depolarization.

Third, we only use a few milliseconds of the VC protocol to provide mechanistic explanations for AP morphological changes — however, there is so much more information embedded in this rich VC data. We believe future work should be directed towards the development of experiments and computational methods that can help decode the ionic current contributors at each time point. The insights drawn from such methods could provide a much more detailed mechanistic explanation for AP morphology and drug effects.

Fourth, ideally we would prefer to conduct these experiments at physiological temperatures, using perforated patch, and without intracellular fluoride. Our attempts to conduct these experiments at physiological temperatures and with perforated patch resulted in very low success rates. With additional cell and experimental optimization, we believe it is possible to address all three of these limitations.

Finally, it would be desirable to test specific ion-channel blockers with the VC protocol. Such a study would help us validate: 1) which currents are present in these cells, and 2) how well the VC protocol isolates currents — based on this study, we know that INa appears to be present in these experiments and IKr is not.

4.5. Conclusion

Automated patch-clamp systems, such as the Nanion Patchliner with Dynamite8, offer promise for increased throughput for studies of iPSC-CM electrophysiology and drug arrhythmia mechanisms. However, the approach has several limitations that must be addressed before it should be accepted broadly as a tool for drug cardiotoxicity screening. We believe there are substantial public health benefits to solving these problems, but it will require a multidisciplinary approach that couples bioengineering and molecular biological methods.

Figure 10: Change in If and IKs segment after drug treatment.

Figure 10:

Open in a new tab

(A) Change in Iout for flecainide- (red) and DMSO-treated (gray) cells during If segment. There was no significant difference between DMSO and flecainide at 3x (p=0.461), 10x (p=0.452), or 20xEFCP (p=0.237). (B) Change in Iout for quinine- (red) and DMSO-treated (gray) cells during If segment. There was no significant difference between DMSO and quinine at 3x (p=.391), 10x (p=.139), but there was at 20xEFCP (p=.0475). (C) Change in Iout for flecainide- (red) and DMSO-treated (gray) cells during IKs segment. There was no significant difference between DMSO and flecainide at 3x (p=0.996), 10x (p=0.433), or 20xEFCP (p=0.766). (D) Change in Iout for quinine- (red) and DMSO-treated (gray) cells during IKs segment. There was a significant difference between DMSO and quinine at 3x (p=.029), 10x (p=.00622), 20xEFCP (p=.0124), and after wash (p=.00895).

Key Points:

  • Human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) offer potential as an in vitro model to study the proarrhythmic potential of drugs, but insights from these cells are often limited by the low throughput of manual patch-clamp.

  • In this study, we use a medium-throughput automated patch-clamp system to acquire action potential (AP) and complex voltage clamp (VC) data from single iPSC-CMs at multiple drug concentrations.

  • A correlation between AP upstroke and INa transients was identified and drug-induced changes in ionic currents found.

  • We also characterize the substantially altered physiology of iPSC-CMs when patched in an automated system, suggesting the need to investigate differences between manual and automated patch experiments.

5.4. Funding

This work was supported by the National Institutes of Health (NIH) National Heart, Lung, and Blood Institute (NHLBI) grants U01HL136297 (to D.C.) and F31HL154655 (to A.C.).

Footnotes

5

Additional Information

The raw data files will be uploaded to a publicly available data repository with a DOI upon acceptance.

5.2

Competing interests

The authors declare that they have no competing interests.

5.1. Data availability

Code to reproduce the figures in this manuscript has been made available on GitHub: https://github.com/Christini-Lab/automated-ipsc-cardiotoxicity.git

References

  1. Ahrens-Nicklas RC & Christini DJ (2009). Anthropomorphizing the mouse cardiac action potential via a novel dynamic clamp method. Biophysical Journal 97, 2684–2692. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Becker N, Horváth A, De Boer T, Fabbri A, Grad C, Fertig N, George M & Obergrussberger A (2020). Automated Dynamic Clamp for Simulation of IK1 in Human Induced Pluripotent Stem Cell–Derived Cardiomyocytes in Real Time Using Patchliner Dynamite8. Current Protocols in Pharmacology 88, 1–23. [DOI] [PubMed] [Google Scholar]
  3. Blinova K, Schocken D, Patel D, Daluwatte C, Vicente J, Wu JC & Strauss DG (2019). Clinical Trial in a Dish: Personalized Stem Cell–Derived Cardiomyocyte Assay Compared With Clinical Trial Results for Two QT-Prolonging Drugs. Clinical and Translational Science 12, 687–697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bloothooft M, Verbruggen B, Seibertz F, van der Heyden MAG, Voigt N & de Boer TP (2024). Recording ten-fold larger IKr conductances with automated patch clamping using equimolar Cs+ solutions. Frontiers in Physiology 15, 1–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Borchard U & Boisten M (1982). Effect of flecainide on action potentials and alternating current-induced arrhythmias in mammalian myocardium. Journal of cardiovascular pharmacology 4, 205–12. [DOI] [PubMed] [Google Scholar]
  6. Clark A, Wei S, Fullerton K, Krogh-Madsen T & Christini D (2024). Single-cell ionic current phenotyping explains stem cell-derived cardiomyocyte action potential morphology. American journal of physiology. Heart and circulatory physiology. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Clark AP, Clerx M, Wei S, Lei CL, de Boer TP, Mirams GR, Christini DJ & Krogh-Madsen T (2023). Leak current, even with gigaohm seals, can cause misinterpretation of stem cell-derived cardiomyocyte action potential recordings, 25(9), euad243. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Clark AP, Wei S, Kalola D, Krogh-Madsen T & Christini DJ (2022). An in silico-in vitro pipeline for drug cardiotoxicity screening identifies ionic pro-arrhythmia mechanisms. British journal of pharmacology 179, 4829–4843. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Clerx M, Collins P, de Lange E & Volders PGA (2016). Myokit: A simple interface to cardiac cellular electrophysiology. Progress in biophysics and molecular biology 120, 100–14. [DOI] [PubMed] [Google Scholar]
  10. Crumb WJ, Vicente J, Johannesen L & Strauss DG (2016). An evaluation of 30 clinical drugs against the comprehensive in vitro proarrhythmia assay (CiPA) proposed ion channel panel. Journal of Pharmacological and Toxicological Methods 81, 251–262. [DOI] [PubMed] [Google Scholar]
  11. Doss MX, Di Diego JM, Goodrow RJ, Wu Y, Cordeiro JM, Nesterenko VV, Barajas-Martínez H, Hu D, Urrutia J, Desai M, Treat JA, Sachinidis A & Antzelevitch C (2012). Maximum diastolic potential of human induced pluripotent stem cell-derived cardiomyocytes depends critically on IKr. PloS one, 7(7), e40288. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Fabbri A, Goversen B, Vos MA, van Veen TA & de Boer TP (2019). Required GK1 to Suppress Automaticity of iPSC-CMs Depends Strongly on IK1 Model Structure. Biophysical Journal 117, 2303–2315. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Goversen B, Becker N, Stoelzle-Feix S, Obergrussberger A, Vos MA, van Veen TA, Fertig N & de Boer TP (2018). A hybrid model for safety pharmacology on an automated patch clamp platform: Using dynamic clamp to join iPSC-derived cardiomyocytes and simulations of Ik1 ion channels in real-time. Frontiers in Physiology 8, 1–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Ishihara K, Sarai N, Asakura K, Noma A & Matsuoka S (2009). Role of Mg2+ block of the inward rectifier K+ current in cardiac repolarization reserve: A quantitative simulation. Journal of Molecular and Cellular Cardiology 47, 76–84. [DOI] [PubMed] [Google Scholar]
  15. Ismaili D, Schulz C, Horváth A, Koivumäki JT, Mika D, Hansen A, Eschenhagen T & Christ T (2023). Human induced pluripotent stem cell-derived cardiomyocytes as an electrophysiological model: Opportunities and challenges-The Hamburg perspective. Frontiers in physiology 14, 1132165. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Johannesen L, Vicente J, Mason JW, Sanabria C, Waite-Labott K, Hong M, Guo P, Lin J, Sørensen JS, Galeotti L, Florian J, Ugander M, Stockbridge N & Strauss DG (2014). Differentiating drug-induced multichannel block on the electrocardiogram: Randomized study of dofetilide, quinidine, ranolazine, and verapamil. Clinical Pharmacology and Therapeutics 96, 549–558. [DOI] [PubMed] [Google Scholar]
  17. Kernik DC, Morotti S, Wu HD, Garg P, Duff HJ, Kurokawa J, Jalife J, Wu JC, Grandi E & Clancy CE (2019). A computational model of induced pluripotent stem-cell derived cardiomyocytes incorporating experimental variability from multiple data sources. Journal of Physiology 597, 4533–4564. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lei CL (2020). Model-Driven Design and Uncertainty Quantification for Cardiac Electrophysiology Experiments. Ph.D. diss., University of Oxford. [Google Scholar]
  19. Lei CL, Clerx M, Gavaghan DJ, Polonchuk L, Mirams GR & Wang K (2019). Rapid Characterization of hERG Channel Kinetics I: Using an Automated High-Throughput System. Biophysical journal 117, 2438–2454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Lei CL, Clerx M, Whittaker DG, Gavaghan DJ, de Boer TP & Mirams GR (2020). Accounting for variability in ion current recordings using a mathematical model of artefacts in voltage-clamp experiments. Philosophical transactions. Series A, Mathematical, physical, and engineering sciences 378, 20190348. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Li W, Luo X, Ulbricht Y & Guan K (2021). Blebbistatin protects iPSC-CMs from hypercontraction and facilitates automated patch-clamp based electrophysiological study. Stem Cell Research 56, 102565. [DOI] [PubMed] [Google Scholar]
  22. Lyra-Leite DM, Gutiérrez-Gutiérrez O, Wang M, Zhou Y, Cyganek L & Burridge PW (2022). A review of protocols for human iPSC culture, cardiac differentiation, subtype-specification, maturation, and direct reprogramming. STAR protocols 3, 101560. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Ma J, Guo L, Fiene SJ, Anson BD, Thomson JA, Kamp TJ, Kolaja KL, Swanson BJ, January CT, Kl K, Bj S & Ct J (2011). High purity human-induced pluripotent stem cell-derived cardiomyocytes : electrophysiological properties of action potentials and ionic currents. Am J Physiol Heart Circ Physiol 301, 2006–2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Marino M, Misuri L & Brogioli D (2014). A new open source software for the calculation of the liquid junction potential between two solutions according to the stationary Nernst-Planck equation. arXiv pp. 1–5. [Google Scholar]
  25. Polonchuk L (2012). Toward a New Gold Standard for Early Safety: Automated Temperature-Controlled hERG Test on the PatchLiner. Frontiers in pharmacology 3, 3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Sager PT, Gintant G, Turner JR, Pettit S & Stockbridge N (2014). Rechanneling the cardiac proarrhythmia safety paradigm: A meeting report from the Cardiac Safety Research Consortium. American Heart Journal 167, 292–300. [DOI] [PubMed] [Google Scholar]
  27. Seibertz F, Rapedius M, Fakuade FE, Tomsits P, Liutkute A, Cyganek L, Becker N, Majumder R, Clauß S, Fertig N & Voigt N (2022). A modern automated patch-clamp approach for high throughput electrophysiology recordings in native cardiomyocytes. Communications biology 5, 969. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Sheldon R, Duff H & Koshman ML (1995). Antiarrhythmic activity of quinine in humans. Circulation 92, 2944–50. [DOI] [PubMed] [Google Scholar]
  29. Tomek J, Bueno-Orovio A, Passini E, Zhou X, Minchole A, Britton O, Bartolucci C, Severi S, Shrier A, Virag L, Varro A & Rodriguez B (2019). Development, calibration, and validation of a novel human ventricular myocyte model in health, disease, and drug block. eLife 8, 1–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. U.S. Food and Drug Administration (2005). ICH S7B Note for Guidance on the Nonclinical Evaluation of the Potential for Delayed Ventricular Repolarization (QT Interval Prolongation) by Human Pharmaceuticals. [PubMed] [Google Scholar]
  31. Verkerk AO & Wilders R (2021). Dynamic Clamp in Electrophysiological Studies on Stem Cell-Derived Cardiomyocytes-Why and How? Journal of cardiovascular pharmacology 77, 267–279. [DOI] [PubMed] [Google Scholar]
  32. Virtanen P, Gommers R, Oliphant TE, Haberland M, Reddy T, Cournapeau D, Burovski E, Peterson P, Weckesser W, Bright J, van der Walt SJ, Brett M, Wilson J, Millman KJ, Mayorov N, Nelson ARJ, Jones E, Kern R, Larson E, Carey CJ, Polat I, Feng Y, Moore EW, VanderPlas J, Laxalde D, Perktold J, Cimrman R, Henriksen I, Quintero EA, Harris CR, Archibald AM, Ribeiro AH, Pedregosa F, van Mulbregt P & SciPy. 1.0 Contributors (2020). {SciPy} 1.0: Fundamental Algorithms for Scientific Computing in Python. Nature Methods 17, 261–272. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Woosley R, Heise C, Gallo T, Tate J, Woosley D & KA R (2021). CredibleMeds QTdrug List. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Code to reproduce the figures in this manuscript has been made available on GitHub: https://github.com/Christini-Lab/automated-ipsc-cardiotoxicity.git

RESOURCES